0xSero/GLM-5.3-569B-W4A16
Partial fine-tune vs cyankiwi/GLM-5.3-AWQ-INT4. Partial change: 7% of the weights changed (mlp), in 75 of 79 layers.
Tensors changed: MLP 639/1090, Attention 0/1118, Embeddings 0/1, Norms 0/364, lm_head 0/1, other 0/1
Partial fine-tune vs cyankiwi/GLM-5.3-AWQ-INT4. Partial change: 7% of the weights changed (mlp), in 75 of 79 layers.
Tensors changed: MLP 639/1090, Attention 0/1118, Embeddings 0/1, Norms 0/364, lm_head 0/1, other 0/1